食用植物油的分类和质量鉴别的模式识别研究

Study on Classification and Quality Discrimination of Edible Vegetable Oils by Pattern Recognition

食品真实性的鉴别是食品质量控制和安全中最重要的工作之一,关系到消费者切身利益和身体健康。研究快速的食品质量鉴别和分类技术对食品产品的安全和消费有重要意义。本文以色泽、水分及挥发物、密度、折光率、酸价、皂化值和过氧化值等理化性质为变量,采用多种模式识别技术对54个食用豆油、合格菜籽油与酸败菜籽油样本进行分类和质量鉴别。主成分分析和聚类分析图显示豆油、合格菜籽油、酸败菜籽油被清楚地区分为三组。偏最小二乘法建立的校正模型对未知样品进行预报,显示出较好的预报能力。结果表明:以所选定的理化性质为变量,上述模式识别技术对类似食用植物油的分类和质量鉴定,是非常有效的。

Determination of food authenticity is one of the most crucial issues in food quality control and safety. It touches the consumers' interests and health. To study on rapid methods for confirming authenticity and classification, in conjunction with greater consumer demands and expectation for safer products, is very important. The objective of this work was to develop a model that would distinguish soybean oils from rapeseed oils, quality rapeseed oils from rancid rapeseed oils according to their type and quality by means of chemical and physical properties with colority, moisture and volatile, density, refractive index, acid value, saponification value and peroxide value as variables. Principal component analysis and cluster analysis plots showed that three different vegetable oils were clustered in distinct groups, while each group could be distinguished clearly. Partial least squares was applied to modeling classes on the basis of the chemical data. The results obtained indicated better performance in terms of classification and prediction for the approaches. These approaches would be useful for primary evaluation of similarity and quality control.